Hi, there, I have two theoretic questions on LDSC. I would deeply appreciate if someone could shed light on this.
How to interpret genetic correlation in the context of phenotypic correlation? For example, if I found the rg between SBP GWAS and DBP GWAS is as high as 0.6, but the phenotypic correlation between SBP and DBP is even higher (r=0.7), does this genetic correlation still mean anything since it could be totally due to phenotypic correlation?
Please see the screenshot below, from the Supplementary Note of the paper "LD Score regression distinguishes confounding from polygenicity in genome-wide association studies". Why X2-statistics equals to N * Beta squared?
To follow up: Is there a more detailed derivation process on equation (1.7) in the appendix?
delta method does not help directly unless you know that is the variance of XijXik
Hi, there, I have two theoretic questions on LDSC. I would deeply appreciate if someone could shed light on this.
How to interpret genetic correlation in the context of phenotypic correlation? For example, if I found the rg between SBP GWAS and DBP GWAS is as high as 0.6, but the phenotypic correlation between SBP and DBP is even higher (r=0.7), does this genetic correlation still mean anything since it could be totally due to phenotypic correlation?
Please see the screenshot below, from the Supplementary Note of the paper "LD Score regression distinguishes confounding from polygenicity in genome-wide association studies". Why X2-statistics equals to N * Beta squared?
Your clarification is greatly appreciated!
Best regards, Jack